OnBenchMark Logo


Jr Data Scientist
Total Views119
back in time
Member since30+ Days ago
 back in time
Contact Details
phone call {{contact.cdata.phone}}
phone call {{contact.cdata.email}}
Candidate Information
  • User Experience
    Experience 3 Year
  • Cost
    Hourly Rate$5
  • availability
  • work from
    Work FromAny
  • check list
    CategoryInformation Technology & Services
  • back in time
    Last Active OnJune 11, 2024
Key Skills
Data ModelingData AnalysisPythonData ExtractionSQLPower BI


● Eye for Blind: Created a computer vision project for visually impaired. Developed a CNN-RNN model with attention mechanism for converting images into descriptive text. Used TensorFlow for model construction. Integrated a text-to-speech API to generate audio output. Explored topics such as CNN-RNN architecture, Attention model with Encoder-Decoder, evaluated the model's performance with Bleu score metric.

● Banking term deposit Prediction: Predictive model using PySpark on Databricks. After preprocessing used Logistic Regression for model building, using areaUnderROC as the evaluator to assess model performance.

● Identifying Entities in Healthcare: Name Entity Recognition (NER) with 91% F1 Score with Part-of-Speech (POS) tagging & CRF Model.

● Automatic Ticket Classification: Non-negative matrix factorization (NMF) under topic modelling and evaluated Naïve Bayes, Decision Tree and Random Forest. Selected Logistic Regression with 93.9% accuracy providing a reliable Natural Language Processing solution.

● Melanoma Detection: Developed a Convolutional Neural Network (CNN) based melanoma detection model with 88% accuracy. Overcame highly imbalanced dataset by utilizing Augmentor to automate image augmentation, leading to a generalized model.

● Gesture Recognition: Explored two architectures CNN & RNN and 3D Convolutional Network (Conv3D). Tried Conv3D, 2D, CNN with LSTM, 2D CNN with GRU and transfer learning MobileNet with GRU. Selected Conv3D architecture as the final model with 95% accuracy.

● House Price Prediction: Used Linear Regression with RFE. Fine-tuned using regularization (Ridge and Lasso). Selected Ridge regression with lowest RMSE and R-squared scores above 90% for both train and test data, achieved by utilizing optimum number of features.

● Telecom Churn Study: Predictive model for customer churn identification. Reduced attribute dimensionality, with PCA followed by fitting ML Models such as Logistic Regression, Random Forest and XGBoost. Selected Logistic Regression with PCA with 94% accuracy.

● Bike Rental Prediction: Multiple Linear Regression Model and refined its accuracy by using p-Values, VIF and RFE.

● Online Grocery Recommender System: Reduce grocery ordering time with accurate product recommendations. Used Association Rules, Collaborative Filtering and Alternating Least Squares (ALS). Deployed the model using Flask and Heroku.

Copyright© Cosette Network Private Limited All Rights Reserved
Submit Query
WhatsApp Icon